David J. Currie

Aim: MaxEnt has been widely used to model species’ geographic distributions as functions of environmental variables and to predict changes in distributions in response to environmental change. Here, we test the predictive ability of MaxEnt models through time by modelling colonizations and extinctions.Location: North America.Methods: Using data for 21 species from the North American Breeding Bird Survey, we first related avian species’ geographic distributions to the spatial variation in environmental conditions. Then, we modelled site‐specific colonizations and extinctions between 1979 and 2009 as functions of MaxEnt‐estimated habitat suitability and neighbourhood occupancy.Results: We found that colonization and extinction probabilities were related to spatial variation in habitat suitability, and to neighbourhood occupancy, in the expected directions. However, change in habitat suitability (which is much smaller through time than through space) is a weak predictor of extinction and worse for colonization. This is because a) for most species and most sites, climatic suitability did not change dramatically between 1979 and 2009, and b) the relationship between colonization or extinction probability and change in climatic variables is very weak (r² = 0.02). Most colonizations and extinctions are apparently unrelated to climate change.Main conclusions: MaxEnt models apparently capture a real effect of habitat suitability on North American bird species’ distributions, but over short and medium time scales, occupancy of neighbouring sites by conspecifics predicts changes in occupancy as well as, or better than changes in climatic habitat suitability, as characterized by MaxEnt. One would not expect species’ distributions to track climate change closely. Prediction of species’ responses to climate change should 1) recognize that the process of colonization and extinction are not equally well predicted by species distribution models and 2) account for the spatial structure of species’ distributions.

Many studies correlate geographic variation of biotic variables (e.g., species ranges, species richness, etc.) with variation in environmental variables (climate, topography, history). Often, the resulting correlations are interpreted as evidence of causal links. However, both the dependent and independent variables in these analyses are strongly spatially structured. Several studies have suggested that spatially structured variables may be significantly correlated with one another despite the absence of a causal link between them. In this study we ask: if two variables are spatially structured, but causally unrelated, how strong is the expected correlation between them? As a specific example, we consider the correlations between broad-scale variation in gamma species richness and climatic variables. Are these correlations likely to be statistical artefacts? To answer these questions, we randomly generated pseudo-climatic variables that have the same range and spatial autocorrelation as temperature and precipitation in the Americas. We related mammal and bird species richness both to the real and the pseudo-climatic variables. We also observed the correlations among pseudo-climate simulations. Correlations among randomly generated, spatially unstructured, variables are very small. In contrast, the median correlations between spatially structured variables are near r2=0.1 – 0.3, and the 95% confidence limits extend to r2=0.6 – 0.7. Viewing this as a null expectation, given spatially structured variables, it is worth nothing that published richness–climate correlations are typically marginally stronger than these values. However, many other published richness–environment correlations would fail this test. Tests of the “predictive ability” of a correlation cannot reliably distinguish correlations due to spatial structure from causal relationships. Our results suggest a three-part update of Tobler’s “First Law of Geography”: #1) Everything in geography that is spatially structured will be collinear. #2) Near things are more related than distant things. #3) The more strongly spatially structured two variables are, the stronger the collinearity between them will be.

A dilemma in ecology: Ecologists aspire to build a discipline to both understand the natural world and to provide society with tools to make responsible decisions about the environment. For both of these purposes, most sciences have, at their core, a set of empirical generalizations that predict the behaviour of important properties of nature (e.g., Newton’s laws of mechanics, Mendeleev’s periodic table, Mendel’s genetics). Ecological science, in contrast, has favoured studies to understand processes (competition, population regulation, etc. – i.e., independent variables) rather than models that predict attributes of nature (dependent variables). Classical reductionist scientific training emphasizes studies of mechanisms under controlled experimental conditions. Yet, inferences about nature from experiments are nearly always unjustifiable extrapolations beyond the experimental conditions. Mechanisms that are statistically detectable in experimental systems may contribute very little to the variation of nature. Studies of ecological processes in isolation may contribute to “expert understanding,” but experts have been shown to be poor predictors of the behaviour of natural systems.A proposed solution: The more relevant, often neglected question is: what factors can statistically account for the observed variance of nature? A more Newtonian approach to ecology would: (a) first, specify the properties of nature (i.e., dependent variables) whose variance is of concern; (b) develop models that statistically capture the variance of those properties in nature; (c) demonstrate that those models can predict independent data; and, only last, (d) experimentally test hypotheses about processes that could give rise to the predictable patterns in nature.Why it matters: Successful disciplines identify specific goals and measure progress toward those goals. Predictive accuracy of properties of nature is a measure of that progress in ecology. Predictive accuracy is the objective evidence of understanding. It is the most useful tool that science can offer society.

Aim: We test macroecological hypotheses (H1: long‐term climate stability; H2: climate seasonality; H3: climate distinctiveness or rarity; and, H4: spatial heterogeneity in contemporary climate, topography or habitat) proposed to explain broad‐scale patterns of total species endemism.Location: Continental areas worldwide and zoogeographical realms.Methods: Using species distribution maps for mammals and amphibians, we calculated five metrics of species endemism, based on inverse and median range size, and range size cut‐offs. We performed multi‐model averaging. We tested the accuracy of fitted models using cross‐validation, comparing observed versus predicted values of endemism among zoogeographical realms.Results: Model averaging showed that species endemism for amphibians and mammals was statistically related to all set of predictors (H1, H2, H4 and richness), except for climate distinctiveness (H3). Effect sizes for spatial heterogeneity were larger and more consistent among zoogeographical realms than the effect sizes of climate stability. The effect of species richness on species endemism varies widely and depends on the metric of endemism and taxonomic group used in the analysis. Cross‐validation across all zoogeographical realms showed that predictions of endemism systematically fail for both taxa. In most cases, low to moderate endemism is predicted reasonably well. However, areas with high numbers of endemic species are systematically under‐predicted.Main conclusions: Our results are not consistent with any of the processes hypothesized to create and maintain global patterns of endemism. Although we found statistically significant relationships, they failed the stronger test of a causal relationship: accurate prediction in independent data. The inconsistent effect of richness in our models suggests that patterns of endemism are not driven by the same variables as total richness. Patterns of endemism have no consistent relationships with climatic stability among zoogeographical realms, so suggestions that endemism reflect climatic stability are likely due to collinearity with other factors, especially spatial heterogeneity.

Executive summary. Through this editorial we seek your support and engagement as authors, readers and reviewers as we take the next steps in developing Frontiers of Biogeography as a leading international journal of biogeography and related subdisciplines. Here we make the case for submitting your next contribution to this journal: affordable, gold libre, open access, with the support of a disciplinarily-informed editorial and review team, which returns benefits to the biogeography community.

Executive summary. Through this editorial we seek your support and engagement as authors, readers and reviewers as we take the next steps in developing Frontiers of Biogeography as a leading international journal of biogeography and related subdisciplines. Here we make the case for submitting your next contribution to this journal: affordable, gold libre, open access, with the support of a disciplinarily-informed editorial and review team, which returns benefits to the biogeography community.

Aim: It is commonly asserted that climate presents the primary constraint on species’ geographic distributions, and that species’ distributions should therefore shift to track changing climate. However, the evidence is surprisingly mixed about the causal link between species’ distributions and climate. Correlations between distributions and climate may be indirect, reflecting influences of other spatially structured habitat variables (e.g., land cover) and/or population processes (e.g., dispersal). Here, we ask if species’ geographic distributions are more strongly related to climate, or to other spatially structured variables.Location: The contiguous United States and southern Canada.Time period: 1990–2000.Major taxa studied: North American passerine birds.Methods: We used the breeding‐season distributions of 19 widespread species of passerine birds whose breeding‐season distributions fall entirely within the area sampled by the North American Breeding Bird Survey. We related these distributions to temperature and precipitation, geographic coordinates, and the degree of occupation of neighboring sites by conspecifics. Two spatial scales were examined: the geographic location of species’ ranges within North America, and site occupancy within species' ranges. We examine these relationships using generalized linear models, structural equation modeling, random forest models and spatial correlograms.Results: On average, geographic coordinates and a model of neighborhood occupancy outperform a simple climatic model. After controlling for geographic coordinates, species occupancy is poorly related to climate. Neighborhood occupancy accounts for the majority of variance captured by geographic coordinates within ranges, and more for the continental placement of ranges. On average, species’ distributions are more strongly and more directly related to spatial coordinates, and neighborhood occupancy, than to climate.Main conclusions: Our results are inconsistent with the hypothesis that climate is the primary, direct determinant of species’ geographic distributions. Rather, other spatially structured factors appear to be stronger determinants of both continental range placement and within‐range distributions of this sample of North American birds.

Aim: Many studies postulate that physiological tolerance of climatic variables imposes the primary limit on species geographical distributions, that tolerances are constant through time, that climate has warmed and that geographical distributions shift to maintain species in their thermal niches when climate changes. However, recent studies present evidence that is inconsistent with each of these propositions. Here we ask: how strongly did avian species entire geographical distributions (as opposed to their latitudinal extremes) in North America track temperature changes between 1979 and 2010?Location: The continental United States (excluding Alaska) and southern Canada.Methods: We examined changes from 1979 to 2010 in the geographical distributions, and the realized temperature niches, of 21 species of passerine birds whose entire breeding ranges fall within the area well sampled by the North American Breeding Bird Survey. We related changes in breeding distributions to the concomitant changes in breeding season temperature.Results: The median temperature increased within the breeding ranges of most, but not all, species. Temperature on the coolest 2.5% of routes increased significantly for only 8 of 21 species. Most species' distributions shifted geographically, but the most frequent shift was westward, not northward. Most species' realized temperature niches changed detectably through time, mainly as a result of changing temperature (versus geographical shifts). Where shifts in geographical distribution occurred, in most cases they did not result in smaller changes in species realized temperature niche than species would have experienced by not moving at all. There is little suggestion of a lagged response to climate change.Main conclusions: We find only slight evidence that the geographical distributions of North American passeriform birds, considered in their entirety (as opposed to their latitudinal extremes), tracked temperature change. Of the factors that have driven recent shifts in the geographical distributions of North American avian species temperature change is probably only a minor one.